Artificial Intelligence Models for the Predictive Detection of Cybersecurity Risks in Digital Educational Environments

Abstract

The consolidation of digital educational environments has increased the interdependence between learning platforms, academic systems, institutional networks, cloud services and connected devices, while simultaneously expanding exposure to cybersecurity risks. In response to this situation, artificial intelligence offers capabilities for processing large volumes of information, recognising patterns, detecting anomalies and anticipating signals associated with potential threats. The analysis examined the artificial intelligence approaches addressed in specialised scientific literature for the predictive detection of cybersecurity risks in digital educational environments, considering their characteristics, applications, variables, capabilities and limitations. A qualitative, descriptive and documentary approach was adopted, using the PRISMA 2020 protocol to select and analyse specialised publications. The findings identified supervised, unsupervised and semi-supervised approaches, deep learning, anomaly detection and predictive techniques. Their scope depends on data quality and representativeness, as well as on the ability of the approaches to generalise and interpret the detected signals.